Errors in Emergency and Trauma Radiology
Visit the link below to download the full version of this book:
https://medipdf.com/product/errors-in-emergency-and-trauma-radiology/
Click Download Now
Foreword
Over the past half century, emergency radiology has come into being as a
subspecialty on its own. Corresponding to a dynamic period in CT technol-
ogy, the evolution of the specialty has been profound. What originally began
as an area of interest for some has morphed into a discipline with a potential
to make a vital difference in the care of acutely ill patients and victims of
trauma.
As a radiologist who trained in the late 1990s, I witnessed firsthand how
imaging made major inroads in the care of the emergency department (ED)
patient. When I started residency in 1995, we routinely performed angiogra-
phy for the evaluation of the potentially injured aorta. Some thought that CT
would never achieve the accuracy needed for this potentially lethal injury.
Over two decades later, it is hard to imagine diagnosing traumatic aortic
injury without CT. With the added reliance on imaging, we all became aware
of potential pitfalls. Of course, we all wished to be sensitive in our diagnoses
but in the ED patient, specificity is equally as important. Chasing an artifact
to exclude aortic injury can potentially be lethal in the setting of a pelvic
fracture or grade 5 liver laceration.
CT for pulmonary embolism (PE) provides another such example. As a
resident, angiography was rarely performed, and ventilation perfusion scin-
tigraphy was the standard for the evaluation of a patient with suspected pul-
monary embolism. CT was not felt to be ready to meet this challenge. Boy,
how times have changed! A night in the ED rarely passes without at least one
PE CT ordered. As with any widely accepted protocol, indication creep
occurs and the number of truly positive studies for PE decreases. Understanding
of the potential errors helps the reading radiologist make sure they find the
unusual PE and prevent overdiagnosis.
In Errors in Emergency and Trauma Radiology, Drs. Patlas, Katz,
Scaglione, and colleagues address the potential errors and pitfalls in the ED
patient. By covering all organ systems, they bring together in one place all of
the ways imaging can mislead us in the care of the ED patient. Specific chap-
ters on select patient populations are also incredibly helpful in avoiding the
traps of imaging in the ED. For the experienced ED reader, this work will
serve as a nice review with creative approaches to reinforce techniques to
improve accuracy. For the general reader, it helps put imaging findings in
context so that the radiologist may make a meaningful difference and provide
effective care in some of our most vulnerable patients.
vii
viii Foreword
Anyone taking call will appreciate Errors in Emergency and Trauma
Radiology as a valuable, concise resource that will help diagnostic accuracy
in the ED. Drs. Patlas, Katz, Scaglione, and colleagues deserve much credit
for bringing these potential errors together in one place. These chapters rep-
resent a compendium of learning in the past half century that will help
increase our value in the next half.
Sanjeev Bhalla, MD
Cardiothoracic Radiology, Emergency Radiology
Mallinckrodt Institute of Radiology
St Louis, MO
USA
Contents
1 Errors in Emergency and Trauma Radiology:
General Principles���������������������������������������������������������������������������� 1
Kate Hames, Michael N. Patlas, Vincent M. Mellnick,
and Douglas S. Katz
2 Mistakes in Imaging Interpretation of Traumatic and
Non-traumatic Brain Emergencies������������������������������������������������ 17
Carlos Torres, Nader Zakhari, Francisco Rivas-Rodriguez,
Angela Guarnizo-Capera, Ashok Srinivasan, and Diego Nunez
3 Errors in Emergency and Trauma Radiology:
C-Spine Imaging������������������������������������������������������������������������������ 35
Sadia R. Qamar, Yuhao Wu, Luck Louis, and Savvas Nicolaou
4 Errors in Imaging of Thoracic Trauma ���������������������������������������� 63
Ashwin Jain, John Lee, David Dreizin, Gene Kim,
and Christina A. LeBedis
5 Errors in Non-traumatic Thoracic Imaging���������������������������������� 83
Daria Manos
6 Errors in Imaging of Abdominal and Pelvic Trauma ������������������ 111
Mariano Scaglione, Ettore Laccetti, Roberto Picascia,
Michele Altiero, Francesca Iacobellis, Mathew Elameer,
and Roberto Grassi
7 Errors in Imaging of Non-traumatic
Abdominal Emergencies������������������������������������������������������������������ 125
Maria Zulfiqar, Vincent M. Mellnick, and Michael N. Patlas
8 Errors in Imaging of the Acute Female Pelvis:
Where Do They Occur and How Can We Reduce Them?����������� 141
Ania Z. Kielar, Shauna Duigenan, and Darcy J. Wolfman
9 Errors in Acute Musculoskeletal Imaging ������������������������������������ 167
Travis J. Hillen, Michael V. Friedman, and Jonathan C. Baker
10 Mistakes in Emergency Imaging of Pregnant Patients���������������� 195
Gabriele Masselli and Martina Derme
11 Errors and Pitfalls in Emergency Pediatric Imaging ������������������ 207
Elka Miller, Gali Shapira-Zaltsberg, Rita Putnins,
and Kristin Udjus
ix
Errors in Emergency and Trauma
Radiology: General Principles 1
Kate Hames, Michael N. Patlas,
Vincent M. Mellnick, and Douglas S. Katz
In 2016, researchers estimated that more than diagnostic-related payments in the USA totalling
251,000 patients die in US hospitals annually as a $38.8 billion [8].
result of preventable errors, ranking medical error A diagnostic error is defined as a medical error
as the third most common cause of death in the related to a missed, incorrect, or delayed diagnosis
USA [1]. Many of these preventable deaths are that is discovered by subsequent findings or tests
due to diagnostic errors. Multiple large autopsy [9, 10]. As medical imaging is central to the over-
studies dating from 1957 [2] describe diagnostic all diagnostic process, it is logical to conclude that
error rates across all medical specialties rang- the incidence of diagnostic error (missed, incor-
ing from anywhere between 5% and 47% [2–7]. rect, and delayed) is attributable, at least in part,
Diagnostic errors in medicine are a major source to radiology-related errors [11]. For example, in a
of patient harm, and result in death more often review of closed malpractice claims in the USA,
than other medical errors including drug-related diagnostic radiology was the sixth more frequent
errors [8]. In addition to affecting patient morbid- specialty involved [12], while approximately three
ity and mortality, diagnostic errors also account out of four malpractice claims against radiologists
for the leading type of paid claims (28.6%) and mention errors in interpretation resulting in missed
the highest proportion of total payments (35.2%) diagnoses [5, 13].
in malpractice lawsuits, with a 25-year sum of Radiology, similar to many other highly com-
plex visual perception-based activities includ-
ing air traffic control or operating nuclear power
K. Hames · M. N. Patlas (*) plants, relies on a sophisticated interplay of
Department of Radiology, McMaster University,
Hamilton, ON, Canada numerous psychophysiological factors and visual
e-mail: [email protected]; perception and is therefore prone to human error
[email protected] [14–17]. Radiological diagnosis also involves
V. M. Mellnick decision-making under conditions of often sig-
Department of Radiology, Mallinckrodt Institute of nificant uncertainty in which the availability
Radiology, Washington University School of of clinical information, prior examinations, or
Medicine, St. Louis, MO, USA
e-mail: [email protected] use of proper technique may be variable [18].
These conditions are amplified in the fast-paced
D. S. Katz
Department of Radiology, NYU Winthrop Hospital, and high-stress environment of emergency and
Mineola, NY, USA trauma centers in which the acuity of poly-
e-mail: [email protected]; trauma patients, involvement of a large multi-
[email protected]
disciplinary team, and the need to make quick
© Springer Nature Switzerland AG 2019 1
M. N. Patlas et al. (eds.), Errors in Emergency and Trauma Radiology,
https://doi.org/10.1007/978-3-030-05548-6_1
2 K. Hames et al.
life-saving decisions all predispose the radiolo- may be either internal (specific to the individual
gist to interpretive error. Under such conditions radiologist) or external (due to larger systemic
of uncertainty, all diagnostic decisions therefore failures). To subdivide these categories further,
have inherent error rates [19]. internal factors include both perceptual and cog-
In the first landmark study of its kind, in 1949, nitive errors. Among internal sources of error,
California radiologist L.H. Garland published an perceptual errors account for approximately
article entitled, “On the Scientific Evaluation of 60–80% of missed or delayed diagnoses in radio-
Diagnostic Procedures,” in which he demonstrated logical interpretation [5, 11, 36–38]. A percep-
a surprising degree of inaccuracy in numerous tual error occurs during the first step of image
clinical, laboratory, and radiological tests [20]. interpretation. For an error to be categorized
Regarding radiological examinations specifically, as a perceptual error, the imaging finding must
Garland discovered a 33% retrospective error rate be deemed sufficiently conspicuous and detect-
among radiologists interpreting positive chest radio- able in retrospect by the initial radiologist or in
graphs and a 2% overcall rate for normal examina- the consensus of his or her peers [11]. As such,
tions [21]. This retrospective experimental error not all subtle or inconspicuous findings that are
rate translates into an error rate of approximately subsequently identified and found to represent a
3–5% when evaluating the prospective interpreta- pathological process would be classified as per-
tion of all examinations during a routine clinical ceptual errors [11]. Considering that the radio-
day [5]. Nearly 70 years later, despite remark- logical error rate has remained stable at 3–5%
able technological advances in medical imaging, for nearly 70 years as noted, it is reasonable to
Garland’s findings on the incidence of radiological assume that every radiologist has committed a
error remain nearly identical. From the 1950s to the perceptual error: a miss that, in retrospect, may
present day, studies have repeatedly demonstrated appear obvious to both the original radiologist
the incidence of diagnostic error in radiology to be and to her or his peers.
approximately 3–5% [17, 19, 22–30]. The psychophysiologic and cognitive pro-
Unlike physical examination findings, radio- cesses by which an obvious abnormality can
logical examinations are now easily accessible simply go unseen when it is so clearly seen
electronic databases which are available for sub- in retrospect have yet to be fully explained to
sequent scrutiny and analysis. Because of the anyone’s satisfaction. Although an increased
accessibility and relative permanence of radio- incidence of perception error may be due to
logical examinations, the extensive collection of other specific risk factors including radiologist
examinations also provides a robust data source fatigue, interruptions, distractions, reading too
from which not only to assess inter- and intra- rapidly, satisfaction of search, or various forms
observer variation, but also to retrospectively of cognitive bias as this chapter will discuss,
detect patterns in errors or discrepancies for most perceptual errors lack a clear identifiable
educational purposes. As dozens of studies have cause. However, studies on radiologist percep-
repeatedly shown, radiological errors follow pre- tual errors from around the world, involving
dictable patterns [5, 14, 18, 22, 30–35]. By ana- radiologists at all levels of training and expe-
lyzing these patterns, individual and system-wide rience and across all modalities, conclude that
measures may be enacted to help prevent similar perceptual errors are not a result of careless-
errors from being made in the future. ness or negligence; rather, perceptual errors
are deemed a consequence of the physiological
processes of human perception and an inherent
1.1 G
eneral Errors in Radiology feature of the complex system in which radiolo-
gists operate [11, 13, 14, 26, 37, 42, 43].
Radiological errors may be categorized in mul- While perceptual errors account for approxi-
tiple different ways [5, 11, 30, 32, 33, 36–42]. In mately 60–80% of interpretive errors, the remain-
the broadest terms, the cause of interpretive error ing 20–40% of internal errors may be classified
1 Errors in Emergency and Trauma Radiology: General Principles 3
as cognitive errors [5, 11, 36–38]. Cognitive 47]. Underreading is the equivalent to a percep-
errors have been defined as “judgment errors” tual miss, where the finding is identifiable but
[5], “faulty reasoning” [22], or “logic fallacies” was overlooked by the first radiologist [30, 42].
[44], in which an abnormality is identified, but its Complacency occurs when a finding is identi-
clinical significance is misinterpreted, resulting fied but is attributed to the wrong cause and not
in an inaccurate diagnosis [11]. Cognitive errors deemed pathological, while faulty reasoning
may be a result of lack of knowledge, faulty occurs when a finding is seen and interpreted
reasoning, or a multitude of cognitive biases. as abnormal but is subsequently attributed to an
Additionally, these biases may be secondary to incorrect etiology [30, 42]. Satisfaction of search
undue influence of previous erroneous reports is another common radiological interpretive error
(known as an alliterative error) or misleading and one that produces nearly as much frustration
clinical information that misdirects the radiologi- in the radiologist as perceptual errors. Satisfaction
cal gaze. However, interpretive errors are more of search is the premature discontinuation of a
likely due to a combination of multiple factors, diagnostic search pattern after a primary, usu-
both intrinsic and extrinsic to the radiologist ally more obvious abnormality is detected [34,
interpreting the imaging examination. 48–51]. Once a single prominent abnormality is
Of the numerous cognitive biases that may identified, the “search for meaning” is satisfied,
influence a radiologist’s interpretive process, four and the interpreter ceases to search for additional,
primary types have been repeatedly identified as usually more subtle abnormalities.
potential causes of diagnostic error: anchoring, In addition to internal factors, there are
framing, availability, and alliterative [11, 31, 44– numerous external factors that also play a sub-
46]. Anchoring bias occurs when the radiologist stantial role in radiological error. These external,
fails to alter his or her initial interpretation despite or system-based, factors include poor or limited
being provided with contrary information [11, radiological technique, lack of access to poten-
31, 44]. Framing bias occurs when the radiologist tially relevant prior imaging, inadequate or mis-
is unduly influenced by the wording or framing directed clinical history, increasing volume and
of the clinical problem, which leads to restricted complexity of cases, staff shortages, constant
diagnostic possibilities [31, 44]. Availability bias interruptions, and reader fatigue, to list just a few
is defined as the propensity to consider a diag- of the more significant external sources poten-
nosis that comes to mind more readily to be the tially contributing to interpretive error [5, 18, 30,
correct diagnosis [11, 31, 44]. This is more likely 32, 42, 44, 52]. The lack of prior imaging exami-
to occur after the radiologist has committed an nations, or the failure to review relevant exami-
interpretive error, which predisposes him or her nations, also contributes to interpretive error [32,
to mistakenly attribute the previously “missed” 42]. Both scenarios suggest that interconnected
diagnoses to a similar finding in a subsequent networks of electronic medical records including
patient [44]. An alliterative error occurs when the radiological examinations, and increased ease of
results from the interpretation of a previous imag- access to such prior exams, would help reduce
ing examination biases the radiologist toward the interpretive error.
same diagnosis when interpreting the current The ever-increasing volume and complex-
examination, which results in a diagnostic error ity of radiological examinations, in addition to
[11, 31, 44]. Another cognitive bias described staff shortages, have led to mounting pressure
by Bruno et al. [11] is the “zebra retreat,” which on radiologists to read more in a shorter period,
occurs when the radiologist resists proposing a which in turn results in longer work hours and
rare diagnosis (despite supportive findings) due mounting reader fatigue, all of which contribute
to the rarity of the diagnosis. to diagnostic error [44, 53–57]. Not surprisingly,
Additional cognitive errors include compla- increasing one’s speed at image interpretation is
cency, faulty reasoning, lack of knowledge on also a source of error. Sokolovskaya et al. [58]
the part of the viewer, and underreading [30, 42, demonstrated that when radiologists interpreted
4 K. Hames et al.
examinations at twice the speed of their baseline, Therefore, it is in the patients’ and the radiolo-
the number of significant errors increased from gists’ best interests to communicate – and docu-
10% to 26.6%. Constant interruptions and multi- ment – urgent findings quickly, and to explicitly
tasking may also result in increased interpretive recommend appropriate additional imaging or
error. Balint et al. [59] studied the number of tele- clinical/laboratory follow-up as necessary.
phone calls on-call radiology residents received
at night, and compared the increased interrup-
tions to the rate of interpretive error (defined as a 1.2 E
rrors in Emergency
resident-attending discordant report). The study and Trauma Radiology
found that in the hour preceding the interpretive
error, a single additional phone call above the The potential for diagnostic error, whether due to
baseline increased the likelihood of a major dis- perceptual errors, cognitive biases, or technical
crepancy by 12% [59]. errors, is further magnified in emergency depart-
One of the most important sources of radio- ments and trauma centers. The fast-paced setting
logical error occurs at the start of the imaging and high-stress environment of emergency and
cycle with the examination requisition and clini- trauma departments create a potential “perfect
cal history. Pinto et al. [40] noted that the study storm” for diagnostic errors: medically unstable
of radiological errors has traditionally been lim- and/or uncooperative patients, insufficient histo-
ited to errors in the radiologist’s report, which ries, multiple concurrent tasks, involvement of a
are frequently taken out of the larger diagnostic large multidisciplinary trauma team, severity and
context, thereby omitting the integral role of the complexity of trauma injuries, quick life-saving
referring physicians. In the majority of studies on decisions, and often junior physicians with less
radiological errors, researchers have found that a experience working after hours when the trauma
relevant clinical history can improve diagnostic volume is typically highest [67–70]. Radiological
accuracy during both the perception and interpre- errors may also be caused by radiologist fatigue
tation phases [46, 60–63]. Loy and Irwig’s [60] and ocular strain from longer work hours, mul-
examination of 16 studies analyzing the accuracy tiple interruptions, lack of prior imaging for
of reports with and without clinical history found comparison, the pressure to read examinations
that providing relevant clinical history improved quickly, and the variable conspicuity of acute
the sensitivity of findings without decreasing abnormalities in difficult-to-image poly-trauma
specificity. Similarly, Leslie et al. [63] found patients. Patients who present to emergency and
that when referring clinicians provided a clini- trauma departments are typically those with more
cal history, radiologists changed 19% of their CT acute injuries, and therefore carry an increased
reports, more than half of which reflected major risk of morbidity and mortality at baseline. As
changes. Providing accurate clinical information such, the diagnostic errors committed in this acute
also ensures that the appropriate radiological setting carry a greater risk of severe complications
examination is performed, and ultimately assists and worse patient outcomes, including death.
the diagnostic workup [44, 46, 64]. Multiple studies evaluating missed injuries
While 40–54% of medical malpractice law- and delayed diagnoses in the emergency set-
suits against radiologists are due to diagnostic ting have been published, with a reported inci-
errors [65], the majority of the remaining legal dence of 1.3–39% [67, 71–77]. Among patients
complaints are due to failure to communicate with missed injuries, 15–22.3% had clinically
the findings in a timely manner, and the failure significant findings [77]. Gruen et al. [67] found
to suggest the next appropriate procedure or that among trauma patients who died from
examination (imaging or otherwise) [47]. Failure recognizable errors, 16% died from delayed
to communicate clinically significant findings operative or angiographic control of an acute
rapidly is the fourth most frequent medical mal- abdominal or pelvic hemorrhage, and 9% died
practice claim made against radiologists [66]. from delayed intervention for on-going intratho-
1 Errors in Emergency and Trauma Radiology: General Principles 5
racic hemorrhage. In autopsy studies involving acquisition, higher spatial resolution, multi-pla-
poly-trauma patients, researchers found that the nar and 3D reformats, and decreased radiation,
primary cause of death was due to severe hemor- have resulted in the increased use of MDCT in
rhage from traumatic bronchopulmonary vessel the emergency setting. The integration of MDCT
injury [78]. Of all the missed injuries in emer- in emergency departments has improved both
gency and trauma centers, Teixeira et al. [72] the speed and accuracy of diagnostic procedures
report that diagnostic errors are responsible for and has led to early detection of clinically sig-
approximately 10–15% of preventable deaths in nificant injuries [77, 87–89], thereby decreasing
trauma center audits. As selective non-operative mortality in trauma patients [90]. With peritoneal
management has become increasingly feasible lavage becoming increasingly obsolete [79, 91],
after abdominopelvic trauma, diagnosis of inju- the diagnosis of poly- trauma injuries, includ-
ries requiring surgery or interventional radiology ing acute arterial hemorrhage, now relies almost
has become more imperative. As such, injuries exclusively on the swift and accurate interpre-
missed on multi-detector computed tomography tation of findings from properly performed CT
(MDCT) have the potential to result in more dire examinations acquired in a timely fashion [83].
consequences. In poly-trauma patients in particular, the pan-
Multiple studies have proven MDCT to be scan CT (head, chest, abdomen, pelvis, and full
superior to both clinical evaluation and diag- spine) is now considered the reference standard
nostic peritoneal lavage for the diagnosis of for the early assessment of acute potentially life-
clinically significant abdominal injuries in poly- threatening injuries.
trauma patients [71, 79–82]. Due to multiple As a key member of the multidisciplinary
factors including decreased consciousness, unre- trauma team, the radiologist not only plays a
liable histories, and distracting injuries, clinical critical role in diagnosing acute life-threatening
examination of trauma patients is frequently injuries but also helps direct the clinical deci-
unreliable [69, 83]. A physical examination sion-making process for surgical or conserva-
of a trauma patient with abdominal injuries is tive management. Therefore, errors in image
only about 60% reliable [69, 84, 85]. As missed acquisition and image interpretation may lead to
abdominal injuries are a well-documented cause suboptimal treatment and potential patient harm.
of increased morbidity and mortality in trauma Radiological errors in the emergency setting fol-
patients [71, 81, 82], early detection of these low predictable patterns, and recognition of these
injuries by CT is crucial to improving patient patterns is crucial to avoiding error and improv-
outcomes. MDCT is also critical to the assess- ing patient outcomes.
ment of head trauma, which is particularly dif-
ficult to assess clinically in many poly-trauma
patients due to decrease levels of consciousness, 1.3 P
erception and Recognition
distracting injuries, and drug and/or alcohol Errors in Emergency
intoxication. Studies have shown that 25% of Radiology
unconscious patients with a serious head injury
have misleading or equivocal clinical findings on Although diagnostic radiology errors are often
examination [69]. In patients with poly-trauma, associated with perception, studies have shown
blunt cerebral-vascular injuries with associated that only 10% of interpretive errors are due to
vertebral and/or carotid injuries in particular are human perception or other nonvisual cues [67,
frequently missed if they are only investigated 72, 92], while approximately 60% of radiologic
with ultrasound, which has been shown to have a errors are caused by poor technique or image
sensitivity of 38.5%, compared to a 100% sensi- quality [93, 94]. One of the most frequent causes
tivity with CT angiography [86]. of diagnostic error in trauma patients is the fail-
Over the past two decades, significant devel- ure to identify fractures on radiographs, which
opments in CT technology, including faster image accounts for 41–80% of interpretive errors
6 K. Hames et al.
in the emergency department [17, 74, 95, 96]. CT examination, thereby exposing the embryo to
Moreover, missed or delayed diagnosis of skel- potentially harmful radiation [104].
etal injuries, particularly fractures of the appen- Other commonly missed injuries in trauma
dicular skeleton, accounts for the majority of patients involve bowel and mesenteric injuries,
malpractice claims against radiologists [74]. The which account for approximately 15–20% of
most commonly missed fractures involve the diagnostic errors [105]. From a clinical perspec-
periarticular regions, shoulder girdle, and feet tive, acute bowel injury often implies surgical
[97]. Approximately 10% of missed fractures exploration, and missed or delayed diagnoses
involve the spine, with the cranio-cervical junc- may significantly increase patient morbidity and
tion (40–50%) and the cervicothoracic junction mortality from sepsis and hemorrhage [106].
being the most common sites of missed injury However, bowel and mesenteric injuries pose a
[97]. While spinal fractures can have significant unique challenge to radiologists, as 9.1–19.4% of
orthopedic and neurological implications, they patients with surgically proven bowel and mes-
may also direct the radiologist to other associ- enteric injuries do not have any identifiable find-
ated injuries. For example, although transverse ings on the preoperative MDCT [107, 108]. More
process fractures are only associated with ver- recent surgical literature has shown an increased
tebral body fractures in 1% of cases, 50% of mortality in patients with a diagnostic delay in
patients with transverse process fractures have bowel injury in as little as 5 h [106]; therefore,
intra-abdominal injuries [98, 99]. Patlas et al. [109] suggest that it may be prudent
Due to the higher sensitivity and specific- to perform a follow-up CT in 6–8 h for patients
ity of CT compared to traditional radiography with clinically suspected bowel injury or new
[100], delayed or missed diagnoses of abdomi- clinical symptoms concerning for bowel injury.
nal and pelvic injuries are less frequent than In addition to recognition errors, interpre-
orthopedic injuries; however, interpretive errors tive errors may also occur when the radiologist
in abdominopelvic injuries carry a greater risk appropriately identifies an abnormality, but mis-
of severe complications due to the potentially takenly attributes it to an incorrect etiology. This
life-threatening nature of solid and hollow organ type of error has been classified as faulty reason-
injury and active hemorrhage. Among solid ing or a misclassification of a true-positive find-
organs, injuries of the liver and spleen each ing [30, 42]. Provenzale and Kranz [41] use the
account for approximately 10–15% of missed or example of venous infarction and dural venous
delayed diagnoses [97]. Although diaphragmatic sinus thrombosis (DST) to illustrate this category
injuries are relatively uncommon and represent of interpretive error. While the radiologist may
only 5% of delayed diagnoses [101], they remain accurately detect the presence of infarction, she
difficult to detect [102]. Radiological suspicion, or he may fail to appreciate a thrombosed corti-
attention to secondary signs, and use of multi- cal vein or dural sinus, and mistakenly interpret
planar reconstructed CT images are crucial for the the finding as an arterial infarct. Similarly, when
correct identification of diaphragmatic injuries. patients with DST receive IV contrast-enhanced
In addition, vascular injuries account for approxi- CT and MRI, the abnormal dural enhancement
mately 5% of delayed diagnoses [97]. In pediat- due to collateral vessels may be mistaken for
ric trauma patients, injuries to the ureteropelvic alternative pathologies such as neurosarcoidosis
junction are overlooked in approximately 50% of or dural metastases [41, 110].
affected patients on the initial image interpreta- Errors also occur when the radiologist mis-
tion [103], which may be due to perceptual error takenly interprets a normal finding as abnor-
as well as technical error if delayed CT images mal, which has been described as overcalling or
are not performed. More than 80% of female false-positive findings [41, 42, 70]. These find-
trauma patients with a previously unknown first- ings may be attributed to poor technique, such
trimester pregnancy are not found to be pregnant as artifact, or anatomical variants mistaken for
during the initial evaluation prior to undergoing pathology. This type of diagnostic error is more
1 Errors in Emergency and Trauma Radiology: General Principles 7
likely to occur among radiology residents or less present with potentially life-threatening inju-
experienced radiologists who both lack experi- ries, radiologists are particularly vulnerable to
ence and who tend to be overly cautious [41]. satisfaction of search errors. In satisfaction of
For example, on CT images, respiratory motion search errors, as previously described, once a
artifact may produce an indistinct gray margin major abnormality is identified, the radiolo-
around the liver, spleen, kidney, abdominal wall, gist may rapidly shorten her or his search time,
or ribs [70]. This linear or halo-like appearance thereby overlooking additional abnormalities
may be mistaken for a subcapsular hematoma or [30]. As Berbaum et al. [51] noted, satisfaction of
even rib fractures [70]. Similarly, cardiac motion search errors are the result of a deliberate trunca-
artifact in the mediastinum may obscure the aor- tion of a search rather than a faulty search pat-
tic root and produce crescentic gray bands within tern. Poly-trauma patients, by definition, present
the ascending aorta, which may be mistaken for with multiple injuries, many of which may be
acute aortic injury. In addition to motion artifact, life-threatening. It is therefore the radiologist’s
anatomical variants such as a splenic cleft may responsibility to quickly and accurately iden-
also be mistaken for a low-grade splenic lacera- tify the most urgent findings that require imme-
tion [70]. Although this category of error may diate surgical or other clinical interventions,
not result in immediate harm, unlike a missed carefully characterize the findings, and directly
acute positive finding, it may result in unneces- communicate critical findings to the appropri-
sary hospital admission for observation [70] and ate clinical team members. When injuries such
unnecessary follow-up examinations, which may as active vascular extravasation, acute aortic
indirectly lead to patient harm. injury, pneumoperitoneum, or massive pneumo-
In contrast to overcalling, under-calling is thorax are identified, the radiologist may focus
another type of diagnostic error that has the on these findings, and inadvertently abbreviate
potential to contribute to patient morbidity and the remainder of the search, thereby overlooking
mortality. Under-calling occurs when the radi- more subtle, but potentially just as clinically sig-
ologist identified an abnormality but dismissed nificant abnormalities.
it as normal or secondary to artifact. While Due to the acuity of patients in the emergency
over-calling may occur more frequently among department and the speed with which clinical
cautious junior radiologists, under-calling may decisions must be made, strong communication
be more common among experienced radiolo- between the radiologist and the treating physi-
gists who are accustomed to seeing artifacts and cian is critical. In many instances, a final written
are therefore seemingly more confident in their report is not sufficient, as the time delay between
interpretations. Provenzale and Kranz [41] sug- the radiologist completing the report and the ER
gest under-calling may occur subconsciously, physician or surgeon reading the report is unpre-
without deliberation about the nature of the dictable. This delay in communication is one of
findings; however, Scaglione et al. [69] suggest the most frequent causes of medical malpractice
these types of errors may occur as a result of claims made against radiologists [66]. In cases
external pressure to reduce the number of false- of acute, life-threatening findings that require
positive interpretations in order to minimize immediate intervention, direct verbal communi-
unnecessary follow-up. It may be reasonable cation between the radiologist and clinician may
to assume these errors may also be a result of avoid delays in treatment and prevent any confu-
lack of knowledge, whereby an abnormality is sion about the severity of injury. Documentation
identified, but because its etiology cannot be of all verbal reports should include the date, time,
confidently deduced, it is erroneously dismissed name of the clinician(s) with whom the radiolo-
as insignificant, thus resulting in a missed or gist discussed the findings, and a detailed account
delayed diagnosis. of what was discussed [111].
In the faced-paced and high-pressure evalu- Another important communication error
ation of poly-trauma patients, many of whom occurs when the radiologist does not expressly
8 K. Hames et al.
communicate her or his recommendations for cian/health-care practitioner and the radiologist
additional or follow-up imaging or other exami- regarding the clinical suspicion of injury.
nations. As discussed previously, the ACR prac-
tice guidelines state that “follow-up or additional
diagnostic studies to confirm the impression 1.4 T
echnical Errors
should be suggested when appropriate” [112]. in Emergency Radiology
Frequently in poly-trauma patients, these rec-
ommendations are made at the time of scanning Although there has been a great deal of research
at the CT console. For example, delayed phases conducted on diagnostic errors associated with
may be added if there is suspicion for ureteral individual perception and cognitive biases, it is
injury, or a CT cystogram may be recommended important to remember that a far greater percent-
in the case of potential bladder injury. However, age (upward of 60%) of radiological errors are
in patients with equivocal findings who require caused by poor technique or image quality [93,
follow-up, it is important that the radiologist rec- 94]. As MDCT has become the reference standard
ommend both the type of follow-up examination for evaluating poly-trauma patients, adherence to
and the timeline in which it should be performed. proper technique and protocol is critical to avoid
This is particularly crucial for suspected bowel inadequate and potentially non-diagnostic exam-
and mesenteric injury, which may not have any inations. When imaging a poly-trauma patient
imaging findings on the initial MDCT scan, with MDCT, it is important that the radiologist
or the findings may be quite subtle [107, 108]. and CT technologist work in close collaboration
However, if bowel injury is suspected, it is imper- with the trauma team to avoid potential errors and
ative the radiologist recommended follow-up in optimize scanning technique. To start, the patient
as little as 6–8 h [109] to avoid potential sepsis should be undressed to ensure no clothing-related
and hemorrhage [106]. artifacts obscure the image, have at least an 18 g
Emergency physicians and associated health- IV to ensure adequate contrast administration,
care practitioners must also communicate clearly have their arms raised above their head (if pos-
with the radiologist and provide an adequate sible in the context of injury) to avoid bony arti-
clinical history to avoid potential missed and fact in the chest and abdomen, have their arms
delayed diagnoses. Without adequate history of down if the area of interest is in the head or neck,
the mechanism of trauma and presenting injuries, and either be cooperative or sedated to minimize
the radiologist cannot protocol the appropriate motion artifact [113].
cross-sectional examination with the necessary Once the patient is properly prepared for
sequences, which predisposes the radiologist to the CT scan, the appropriate protocol must be
both perceptual and technical diagnostic errors. selected based on the clinical history and mecha-
Scaglione et al. [69] stated that approximately nism of injury. While in some patients this may
40% of the patients with delayed diagnoses are include a full-body scan (head, chest, abdomen,
due to clinical survey oversight. More specifi- pelvis, and spine) with arterial and portal venous
cally, an incomplete history has been shown to phases, other patients may require more tailored
result in a 10% likelihood of delayed diagnosis approaches with additional phases of imaging.
[73]. Additional studies have found that 15% of This includes patients with acute hemorrhage
delayed diagnoses are due to the failure of the who may require multiphasic imaging to accu-
clinician to order appropriate imaging of the rately characterize the source of hemorrhage.
region of injury identified during clinical exami- At the minimum, a CT angiogram of the chest,
nation [73]. Obtaining an adequate history from abdomen, and pelvis is recommended to detect
a trauma patient is notoriously difficult, as noted the source of acute arterial hemorrhage [114–
[97]. However, appropriate imaging and inter- 117], although non-contrast and/or delayed CT
pretation can only be accomplished if there is acquisitions may be useful to better characterize
clear communication between the treating physi- the source of bleeding. However, not all sources
1 Errors in Emergency and Trauma Radiology: General Principles 9
of hemorrhage are arterial; therefore, it is impera- Coronal and sagittal reconstructions are particu-
tive to attempt to identify whether the source of larly helpful for localizing the source of any acute
extravasation is due to either an arterial or venous hemorrhage [17], characterizing spinal fractures,
injury, which will help direct interventional and assessing bowel and mesenteric injury [127],
surgical management as necessary [69, 118]. and identifying diaphragmatic injuries which
After the biphasic (arterial and portal venous) are notoriously difficult to detect only on axial
examination has been performed, the radiologist images [102] and are therefore easily missed.
may detect potential renal, ureteral, or bladder
injuries requiring additional phase images. If
the emergency radiologist is not at the scanner 1.5 S
olutions and Prevention
at the time of CT image acquisition to direct fur- of Radiological Errors
ther imaging, multiple injuries only detectable
on delayed phases may potentially be missed. When considering potential solutions to prevent-
For example, if there is suspicion for renal or ing diagnostic error in radiology, it is important
ureteral injury, a delayed excretory phase is rec- to consider both person-centered and system-
ommended at an 8–12-min delay. If there is sus- based solutions. However, care must be taken
picion for bladder injury, an MDCT cystogram when defining a person-centered approach, as
should be performed to assess the extent of injury the aim is not to focus on an individual who has
and to characterize if it is intra- or extraperito- committed an error and thereby subject him or
neal or both, which will dictate surgical man- her to blame, shame, or disciplinary action [18].
agement [119]. Delayed CT images also help to Such targeted shame-based approaches have
further characterize solid visceral organ injuries been proven counterproductive and ill-suited to
that involve the vasculature and which may also the health-care domain [128]. Instead, the focus
require surgical or urgent interventional manage- should be on the larger forces that have created
ment [120–123]. the conditions for the error to occur. This is not
Appropriate MDCT technique is also crucial to say that solutions such as improved educa-
in the evaluation of cerebral trauma. As blunt tion cannot be directed at particular individuals,
cerebral-vascular injuries are frequently underdi- specifically radiologists-in-training and junior
agnosed in poly-trauma patients [86], full evalu- attendings; rather, continuing education and
ation with a CT angiogram of the carotid and adherence to standards of care should be directed
vertebral arteries as well as of the circle of Willis toward all members of the radiology department.
may help avoid missed or delayed diagnoses of Education on radiological errors, includ-
vascular injury and potentially prevent neuro- ing cognitive biases and the propensity to com-
logical complications. With advances in MDCT mit satisfaction of search errors, may help raise
techniques, a full-body CT angiogram from the awareness of the common “error traps” [18] and
circle of Willis to the pelvis is possible and has thereby decrease the incidence of these errors.
been advocated in patients with severe poly- For example, if awareness is raised about allit-
trauma [124–126]. In order to maximize image erative bias, whereby reading the previous report
quality and prevent missed or delayed diagno- may unduly influence the interpretation of the
ses, appropriate MDCT protocols must be used. current examination, radiologists may choose to
This includes important follow-up examinations avoid consulting the prior report until they have
for patients with equivocal findings on the initial completed their own evaluation of the current
MDCT, such as those with potential bowel or examination. To decrease satisfaction of search
mesenteric injury, as well as in patients with new errors, education on complete search strategies
or worsening symptoms. and common mechanism-based multiple injury
Once all of necessary phases of an MDCT patters may prove beneficial [127]. For example,
examination have been obtained, multi-planar knowledge of trauma injury “packages” (such as
reformatted images must also be evaluated. right-sided injuries or left-sided injuries) may
10 K. Hames et al.
help focus the radiologist’s attention on organ CT scanner at the time of image acquisition in
systems and adjacent structures most likely to order to assess the need for delayed imaging or a
be involved in particular mechanisms of injury, CT cystogram.
thereby decreasing the potential for missed As fatigue and ocular strain have also been
or delayed diagnoses. Additionally, the use of proven to contribute to diagnostic errors [56],
checklists or dictation templates (especially for optimizing ergonomics, encouraging breaks,
residents and junior attendings) may reduce diag- and promoting physical activity whenever pos-
nostic error by promoting a more systematic and sible may prove beneficial in reducing error
complete search process [127]. rates [130]. Optimal lighting and individual-
Intradepartmental and multidisciplinary meet- ized ergonomic settings of PACS stations may
ings focusing on clinical and radiological diag- reduce physical stressors and improve the read-
nostic errors may also prove beneficial from an ing experience, which may potentially decrease
educational standpoint. However, for such meet- diagnostic error [44]. Decreasing the number
ings to be productive and to have a positive learn- of interruptions may also prove beneficial, as
ing outcome, the culture of the meeting must not disrupting radiologists’ focus during image
be one of blame. Fitzgerald [18] noted in 2001 interpretation has been shown to result in inter-
that the culture at that time was to embarrass and pretive errors [59]. For example, Rosenkrantz
shame the radiologist who committed the error. et al. [131] found that the introduction of
This approach has the potential to undermine the reading room coordinators to assist radiolo-
educational value and instead foster a culture of gists with phone calls and other administrative
fear and animosity. Radiological error/quality tasks significantly decreased interruptions and
assurance meetings (or morbidity and mortality improved radiologists’ workflow efficiency.
rounds) may be more beneficial if they are con- Implementing similar programs throughout
ducted according to the principles outlined by radiology departments may also help to reduce
Chandy et al. [129]: a confidential reporting sys- diagnostic errors.
tem, anonymous presentation, written reports by Recent advancements in artificial intelli-
peers at the meeting, and consensus adjudication gence (AI) and machine learning (ML) algo-
on the presence or absence of error. Encouraging rithm also promise to streamline the data mining
radiologists to share their diagnostic misses and organizational tasks that often detract from
and mistakes with others in a supportive learn- radiological interpretation of examinations
ing environment may not only help others avoid [132–135]. For example, Thrall et al. [133]
similar errors but may also lead to greater self- argued that, more than improve diagnostic accu-
awareness of one’s own search errors and cogni- racy, AI can be applied to numerous practical
tive biases, thereby decreasing diagnostic errors issues that radiologists encounter on a daily
overall. basis: optimizing work lists to prioritize cases,
At the system level, it is important that all pre-analyzing cases in high-volume settings
equipment is functioning optimally in order to help eliminate observer fatigue, extracting
to maximize the quality of image production. information from images not readily apparent
Standardized MDCT protocols are also impor- to the human eye, and improving the quality of
tant, particularly in the evaluation of poly-trauma reconstructed images [133]. The application of
patients. Depending on the mechanism of injury AI may also assist with the timeliness of image
and the clinical suspicion, whole-body MDCT interpretation and communication of urgent
protocols including angiographic examinations findings. If algorithms can be used to prescreen
from the circle of Willis to the pelvis may prove examinations rapidly and detect urgent find-
beneficial in detecting otherwise occult inju- ings such as pulmonary emboli, pneumotho-
ries [124–126]. The radiologist should also be rax, pneumoperitoneum, or other potentially
encouraged to be present, when possible, at the life-threatening conditions, the program could
